U.S. v. ATKINS

RANDALL, Circuit Judge:

Defendant appeals pro se the district court's order denying his motion to correct an illegal sentence. Defendant filed the motion after he pled guilty in the Western District of Texas to one count of conspiracy to manufacture amphetamine and after he received, for his plea, a four-year sentence. In his motion to the district court, defendant explained that just prior to entering his plea in Texas, he pled guilty and received a four-year sentence in Oklahoma for conspiring to manufacture phenylacetone and amphetamine. According to defendant, the conspiracy made the subject of the Texas indictment was the same conspiracy on which the Oklahoma indictment focused. Since he is already serving time in Oklahoma for that conspiracy, defendant argued, his Texas sentence subjects him to double jeopardy in violation of the United States Constitution and must be vacated — regardless of his guilty plea. The district court disagreed; it determined from the record before it, and without conducting an evidentiary hearing, that the Texas and Oklahoma indictments charged two separate conspiracies. We cannot, however, agree with the district court that the record in this case is so clear. Because a more detailed analysis of the two conspiracy charges against defendant is needed before his double jeopardy claim can be resolved, we therefore return his claim to the district court for an evidentiary hearing.

I.

FACTS AND PROCEEDINGS BELOW

Richard Lee Atkins ("Atkins") was indicted on June 7, 1983, and charged with five separate violations of the Controlled Substances Act, 21 U.S.C. § 801 et seq. Count one of the indictment charged:1

The count continued by alleging several overt acts which the defendants were to have committed in furtherance of the conspiracy, and concluded that by conspiring to manufacture phenylacetone and amphetamine, the defendants violated Title 21, section 846 of the United States Code. The indictment was issued by a grand jury in the Western District of Oklahoma ("Oklahoma indictment"). Seven weeks later, a grand jury in the Western District of Texas issued a two count indictment against Atkins, also charging him with Controlled Substances Act violations ("Texas indictment"). Count one of the Texas indictment charged:2

No overt acts were alleged in the Texas indictment.

With two indictments and seven counts pending against him, Atkins decided to plea bargain. In Oklahoma, Atkins agreed to plead guilty to counts one and three of the Oklahoma indictment in exchange for the government's dismissal of counts two, four, and five. On September 9, 1983, therefore, Atkins was given four-year consecutive sentences on each of counts one and three. After being sentenced in Oklahoma, Atkins was transferred to Texas where he plea bargained the Texas indictment. This time, Atkins pled guilty to count one in exchange for dismissal of count two. Before accepting Atkins' plea at his Rule 11 hearing, the district court asked the government what it was prepared to prove against Atkins. The government responded that it was prepared to prove that:

On November 9, 1983, Atkins was sentenced to four years on his plea to the Texas indictment; the district court imposed the sentence to run consecutively to the sentences imposed on Atkins in Oklahoma.

Atkins took no appeal from either plea. Instead, two and one-half years later, Atkins filed in Texas a motion to correct an illegal sentence pursuant to Rule 35 of the Federal Rules of Criminal Procedure. In his motion, Atkins argued that the sentence he received as a result of the Texas indictment and subsequent plea bargain placed him in double jeopardy and, therefore, violated the fifth amendment to the United States Constitution. As Atkins explained it, he was sentenced in Texas for the substantive offense of conspiracy to manufacture amphetamine. The first of the two offenses for which he was earlier sentenced in Oklahoma, however, was also a conspiracy offense — conspiracy to manufacture amphetamine and phenylacetone.3 According to Atkins, a comparison of the facts underlying the conspiracy charge in the Texas indictment and the facts underlying the conspiracy charge in the Oklahoma indictment demonstrates that both indictments focused on the same conspiracy. To support his contention that the two charged conspiracies were in fact only one actual conspiracy, Atkins appended to his motion investigative documents prepared by the government in connection with both the Texas and Oklahoma indictments.

The most informative document from the Texas investigation is a report of investigation filed by Special Agent to the Drug Enforcement Administration ("DEA"), Fred E. Moore ("Moore"). Moore's report details information he obtained from undercover DEA Special Agent John E. Coonce ("Coonce") and from local police investigators. Together, this information suggests a story which unfolded on March 30, 1983 in a house in Odessa, Texas. The report indicates that on that date, Atkins was manufacturing amphetamine at an Odessa, Texas residence that belonged to Kim Dansby ("Dansby"). The report explains that after manufacturing approximately twelve ounces of amphetamine, Atkins was in the process of disposing of twelve liters of ether when he received a phone call from some friends. The friends were supposed to be out selling the recently manufactured amphetamine; however, their car had broken down and, at the time of the phone call, they were stranded on a main street in Odessa which was heavily patrolled by the police. Atkins promptly abandoned his attempt to dispose of the ether, and left the house to rescue his friends.

When he and his friends returned later, however, they discovered that an explosion had occurred in their absence. The report speculates that after Atkins left, Dansby decided to dispose of the ether herself. Somehow, the ether was ignited and the house exploded, killing Dansby and badly burning her six-year-old daughter who was asleep in a back room. When Atkins and his friends discovered the explosion, they retreated in a pickup truck which was parked behind the house. Just who Atkins' "friends" were is not disclosed in the report. However, immediately after the explosion, the police noted the license plate of the pickup truck; the police later determined that the truck was owned by Kenneth Gill. Approximately one week after the explosion, the report continues, Atkins purchased new chemicals and glassware — apparently to replace those lost in the explosion. These supplies were purchased from Coonce at the Metroplex Chemical Company, which is located in Dallas, Texas. On April 12, 1983, Atkins bought additional supplies from Coonce. This time when he made his purchases, Atkins was accompanied by Leiann Lynda Gill, the wife of Kenneth Gill. According to the reports, Leiann Gill told Coonce that she was helping Atkins manufacture the amphetamine. The report suggests that Moore and another special agent intended to follow Atkins as he left the Metroplex Chemical Company on April 12. However, the report concludes without disclosing what occurred after Atkins made his last purchases. At this point, two documents from the Oklahoma investigation become helpful.

The first document is an affidavit by an unidentified DEA Special Agent ("the Agent") containing information which the Agent obtained from an unidentified reliable source ("the source") about Atkins' "clandestine manufacture of Phenylacetone and Amphetamine." The affidavit explains that the source had been providing the Agent with information about Atkins since April, 1982. From the affidavit, it appears that the source was based in Dallas. On April 6, 1982, the source learned from Atkins about the explosion and Dansby's death. The Agent subsequently corroborated with the Texas DEA Task Force Atkins' version of the explosion. On April 7, 1983, the source observed Atkins purchase chemicals and glassware at the Metroplex Chemical Company in Dallas. On April 12, 1983, the source again spoke to Atkins in Dallas. This time he learned that Atkins had just finished manufacturing phenylacetone in a barn located behind a residence in Oklahoma. Atkins told the source that he intended to start making amphetamine that evening at the same location. Based on this information, the Agent and several other DEA agents followed Atkins from Dallas to a residence in Choctaw, Oklahoma. At this point, the affidavit ends. However, the criminal complaint filed against Atkins in Oklahoma on April 13, 1983 completes the story.

According to the complaint, in the early hours of April 13, the DEA agents obtained a warrant to search the Oklahoma residence and its "out buildings." While executing the warrant, the agents found Leiann Lynda Gill, Kenneth Kanole Gill, and Roger Ray Helvy in the barn. They also found phenylacetone. During the search of the residence, the agents found Atkins, John Dwayne Snyder, and a woman named Teresa Louise Anfleger. It was later discovered that Snyder owned both the property and the car Atkins drove from Dallas to Oklahoma. The complaint concludes by listing as material witnesses to the charge two DEA special agents: Robert N. Havens and Coonce. Both special agents were based in Dallas.

On the basis of the indictments and the investigative information disclosed above, Atkins asked the district court for relief on double jeopardy grounds from his Texas sentence. In a one page opinion, and without conducting an evidentiary hearing, the district court denied Atkins' request. As reflected in the opinion, the court concluded that "[t]he Defendant's allegations do not fit the five-prong test as set out in United States v. Marable, 578 F.2d 151 (5th Cir.1978). It is quite clear that the conspiracies the Defendant was involved in were two separate conspiracies, and thus the Court's sentence does not constitute double jeopardy." From this ruling, Atkins now appeals. He argues to us that the district court erred in refusing to find that he has been subjected to double jeopardy by being sentenced twice for participating in one conspiracy. In the alternative, Atkins urges that — at the least — the evidence he presented to the district court entitled him to a hearing at which to prove his claim. In response, the government argues that the district court correctly concluded that Atkins pled guilty to two separate conspiracies. In addition, it urges that no hearing was necessary, given the evidence Atkins presented, to determine that Atkins' pleas did not subject him to double jeopardy. After reviewing the question, we conclude that Atkins has not adequately proved the substance of his double jeopardy claim at this point. However, we must agree with Atkins that, on the record before us, the district court should have given him a chance to make his case at an evidentiary hearing.

II.

ANALYSIS

A. Jurisdiction

While the substantive issue in this case concerns principles of double jeopardy, we must first briefly look at the question of our jurisdiction to hear this appeal. As the vehicle for asserting his double jeopardy claim, Atkins chose a motion to correct an illegal sentence under Rule 35(a) of the Federal Rules of Criminal Procedure. Rule 35(a) permits such a motion to be made "at any time." See Fed.R.Crim.P. 35. But once the district court denies a Rule 35(a) motion to correct an illegal sentence, the Federal Rules of Appellate Procedure permit a defendant only ten days in which to file a notice of appeal. See Fed.R.App.P. 4(b); United States v. Babineau, 795 F.2d 518, 519 (5th Cir.1986). In his brief, Atkins admits that he failed to file his notice within the prescribed time period.4 In this case, however, his failure is not fatal either to his claim or our jurisdiction. We have held that in this situation, it is proper for us to treat a prisoner's motion as a habeas petition brought pursuant to section 2255 of Title 28. See Babineau, 795 F.2d at 519; United States v. Santora, 711 F.2d 41, 42 (5th Cir.1983). We therefore elect to construe Atkins' Rule 35 pleading as a request for habeas relief. So construed, Atkins notice of appeal — which was filed seventeen days after the court entered its order denying him relief — is timely. See Fed.R.App.P. 4(a). Consequently, we turn now to Atkins' double jeopardy claim.

B. Atkins' Double Jeopardy Claim

1. Questions Presented

We begin, as always, by setting forth the standard of appellate review which governs our consideration of the district court's judgment. Without conducting a hearing, the district court rejected Atkins' double jeopardy claim on the merits. This single ruling, however, actually encompasses two separate conclusions. First, by ruling against Atkins on the merits of his claim, the court determined that Atkins did not prove that he was subjected to double jeopardy. Second, by failing to conduct a hearing on Atkins' claim, the district court also necessarily concluded that "the motion and the files and records of the case conclusively show[ed] that the prisoner is entitled to no relief." See 28 U.S.C. § 2255; accord Santora, 711 F.2d at 42; United States v. Bondurant, 689 F.2d 1246, 1251 (5th Cir.1982). Double jeopardy is, of course, a question of law. Baker v. Metcalfe, 633 F.2d 1198, 1201 (5th Cir. Unit A Jan. 1981), cert. denied, 451 U.S. 974, 101 S.Ct. 2055, 68 L.Ed.2d 354 (1981). Consequently, we must subject the district court's first ruling to de novo review. Similarly, as we have done before, we must independently examine the motion, files, and records of this case to determine the appropriateness of the district court's second ruling. See Santora, 711 F.2d at 42-43; Bondurant, 689 F.2d at 1251.

To make two such definitive rulings required the district court — and, consequently, now require us — to explore two separate questions: first, what the substantive evidence must show in order to prove a double jeopardy claim in a drug conspiracy case and second, from what sources the substantive evidence must be drawn before a court deciding a double jeopardy claim can consider that evidence. In a sense, these questions are not new to us. As the district court realized, we developed a standard which answers the first question in United States v. Marable, 578 F.2d 151 (5th Cir.1978). In Marable, we recognized that a claim of double jeopardy is proved by demonstrating that the two offenses charged are, both in law and fact, the same offense. Id. at 153; accord United States v. Henry, 661 F.2d 894, 896 (5th Cir. Unit B Nov. 1981), cert. denied, 455 U.S. 992, 102 S.Ct. 1619, 71 L.Ed.2d 853 (1982). However, when charges of conspiracy are the focus of the demonstration, we explained in Marable, a court must not forget that "[t]he gist of the crime of conspiracy and the characteristic which defines its breadth is the unlawful agreement." Id. at 153. Therefore, we acknowledged that "the precise nature and extent of the conspiracy must be determined by reference to the agreement which embraces and defines its objects." Id. (quoting Braverman v. United States, 317 U.S. 49, 53, 63 S.Ct. 99, 101, 87 L.Ed. 23 (1942)).

Focusing on the agreement, however, did not simplify the task of defining the scope of a conspiracy. By their nature, those agreements which drug conspiracy statutes forbid are both hidden and mercurial. Consequently, as an aid for a court attempting to define how many agreements and, therefore, conspiracies are involved in a given case, we stressed in Marable the appropriateness of focusing on five discrete factors:

Marable, 578 F.2d at 154. We required that these factors be reviewed in tandem for a reason. A single agreement can have as its object several different actions or activities. Moreover, the object of the agreement can be achieved over a period of days, weeks, or months. And the people involved in executing the agreement can change from time to time. In none of these situations, however, can a person be punished under one drug conspiracy statute for more than just the single agreement. See, e.g., Braverman, 317 U.S. at 53, 63 S.Ct. at 101 (holding that one agreement cannot be taken to be several agreements and, therefore, several conspiracies because the agreement envisages the violation of several statutes rather than one); United States v. Nichols, 741 F.2d 767, 771 (5th Cir.1984), cert. denied, 469 U.S. 1214, 105 S.Ct. 1186, 84 L.Ed.2d 333 (1985) (fact that crimes are committed at different times not enough to change one agreement into two); United States v. Winship, 724 F.2d 1116, 1127 (5th Cir.1984) (fact that separate controlled substances were involved not enough to change one agreement into two); United States v. Kalish, 690 F.2d 1144, 1151 (5th Cir.1982), cert. denied, 459 U.S. 1108, 103 S.Ct. 735, 74 L.Ed.2d 958 (1983) (fact that crimes were committed at different times and that new people were added to the scheme not enough to change one agreement into two). When applying the Marable analysis, therefore, a court must remember that any one factor taken in isolation can suggest two agreements when only one actually exists. Consequently, a court must also recognize that even though each factor by itself speaks to the existence of an agreement, no one factor definitively suggests the agreement's true dimensions. The combination of the factors, however, is much more telling — like a jigsaw puzzle which gives a complete picture only when all its pieces are viewed together in their proper relation to each other, we have found that the juxtaposition of these five factors effectively paints the contours of the agreement, the scope of the conspiracy.

We emphasize why the Marable factors must be evaluated in tandem to prove a point — when the underlying offenses to a double jeopardy claim are both drug conspiracy charges, we have required a court, when reaching a decision on the merits, to carefully evaluate the claim by piecing together the particular facts of the particular case before it. The quest has always been to determine whether "there emerges ... two discrete patterns of activity [or] a single design with the events most important in each case appearing at crucial and common junctures." See Marable, 578 F.2d at 156. A review of our cases which have applied the Marable factors discloses how focused we have been on this fact-sensitive and claim-specific method of examining drug conspiracy charges for double jeopardy problems.5 Moreover, understanding our method of inquiry in these cases emphasizes how important our second question is: from what sources may a court draw its "facts" when attempting to give content to the five Marable factors? Because our inquiry is so fact-specific, the answer to this question — which tells a court exactly which pieces of information it may plug into the Marable paradigm — can in a real sense decide the outcome of a double jeopardy claim.6

Again, this important second question is not new to us. In Marable, we recognize that "by the nature of the crime, the precise bounds of a single conspiracy seldom will be clear from the indictment alone." Marable, 578 F.2d at 153. We found that this is especially true when the alleged conspiracies both arise under section 846, because "a Section 846 indictment is sufficient if it alleges a conspiracy to distribute drugs, the time during which the conspiracy was operative and the statute allegedly violated, even if it fails to allege or prove any specific overt act in furtherance of the conspiracy." Id. at 154. Consequently, we concluded that "because a Section 846 indictment may be tightly drawn, the court must look to the record to determine whether constituent elements of the two conspiracies charged indicate that the government has twice placed the defendant in jeopardy." Id. In Marable, the "record" to which we turned to supplement the indictment consisted of the evidence used against the defendant at his two conspiracy trials. Our later cases have also reviewed trial testimony. See Levy, 803 F.2d at 1395; Winship, 724 F.2d at 1126; Kalish, 690 F.2d at 1147; Tammaro, 636 F.2d at 102. Moreover, where it is available, these cases have also looked to evidence — ranging from live testimony to DEA investigative documents — submitted at pretrial hearings. See Kalish, 690 F.2d at 1149-50 (signed pretrial statements and live testimony); Henry, 661 F.2d at 896 (transcripts of telephone calls recorded by the DEA and assorted business records); Futch, 637 F.2d at 390-91 (live testimony). None of these cases has suggested any sort of limitation on the source to which a court may turn for facts.

2. The Procedural and Substantive Standards To Be Applied To Atkins' Claim

As we said before, therefore, neither of the two questions we must answer to determine Atkins' double jeopardy claim is new to us. What is new, however, is the context in which these questions have arisen. Marable and its progeny involved claims of double jeopardy for which evidence was developed either before or during trial. In none of those cases had the defendant admitted, by way of a guilty plea, to committing the crime which he was contesting. The case before us today, therefore, differs from our previous cases in a noteworthy respect: Here, the double jeopardy issue arose in a petition for habeas relief after Atkins had pled guilty to both conspiracy charges. The question before us, then, is whether this new context requires new answers to the questions which we decided in Marable. After careful consideration, we conclude that, to a limited extent, it does.

We cannot determine whether new answers are necessary without first understanding why the procedural status of Atkins' claim is important. Because Atkins pled guilty to both indictments before questioning whether they charged the same crime, there is neither pretrial hearing testimony to look to nor trial testimony to review.7 Instead, the record in this case is limited to the indictments themselves and the sketchy facts adduced at the Rule 11 hearing where Atkins' Texas plea was accepted. By pleading guilty, therefore, Atkins foreclosed development at the time of the incident of additional facts which would either prove or disprove his claim. Because of his actions, our — or any court's — ability to review his double jeopardy claim is made more complicated. The lack of a record, the passage of time, and the effect on memory which attends the passage of time, combine to complicate our review. We believe that the law requires Atkins to shoulder the responsibility for the complications which he has created by choosing to raise the issue as he has. At the same time, however, we cannot lose sight of the overriding question which faces us, with all its constitutional implications: Has Atkins, in fact, been subjected to double jeopardy? With this understanding of the importance of the context in which Atkins' claim arises, we turn our attention to the task at hand — evaluating the usefulness of the answers found in Marable to claims which arise in this context.

a. The Burden of Proof

We begin, first of all, by recognizing that the habeas context changes at least one procedural rule which has accompanied the application of the Marable factors. In United States v. Stricklin, 591 F.2d 1112, 1117-18 (5th Cir.), cert. denied, 444 U.S. 963, 100 S.Ct. 449, 62 L.Ed.2d 375 (1979), we determined in a drug conspiracy case that, although the defendant bears the initial burden of proving a "prima facie" nonfrivolous double jeopardy claim when he raises the claim before trial, the government bears the ultimate burden of proving that the indictments charged separate crimes. This burden of proof allocation does not, of course, remain the same when the petitioner chooses to first assert his double jeopardy claim in a habeas proceeding. As we have noted before, the "substantial difference between direct and collateral attacks" is enough to justify a different allocation of the burden of proof. Bruce v. Estelle, 536 F.2d 1051, 1058-59 (5th Cir.1976), cert. denied, 429 U.S. 1053, 97 S.Ct. 767, 50 L.Ed.2d 770 (1977). As is usually the case in habeas proceedings, therefore, a petitioner who chooses to raise his double jeopardy claim at this point has the burden of proving by a preponderance of the evidence that his constitutional rights have been violated. See Hawk v. Olson, 326 U.S. 271, 279, 66 S.Ct. 116, 120, 90 L.Ed. 61 (1945); Johnson v. Zerbst, 304 U.S. 458, 468, 58 S.Ct. 1019, 1024, 82 L.Ed. 1461 (1938); Bruce, 536 F.2d at 1058.

b. Substantive Law on Conspiracy

Turning to Marable itself, we conclude that there is no justification for tampering with its five-factor analysis for determining the number of drug conspiracies with which the government has charged a defendant. As the Supreme Court explained in Brown v. Ohio, 432 U.S. 161, 97 S.Ct. 2221, 53 L.Ed.2d 187 (1977), the double jeopardy clause provides three categories of protection. It "protects against a second prosecution for the same offense after acquittal. It protects against a second prosecution for the same offense after conviction. And it protects against multiple punishments for the same offense." Id. at 165, 97 S.Ct. at 2225 (quoting North Carolina v. Pearce, 395 U.S. 711, 717, 89 S.Ct. 2072, 23 L.Ed.2d 656 (1969)). Therefore, whether a double jeopardy claim is examined before trial, during trial, or in a habeas petition, the examining court's focus will be on the same point — the nature of the offense. As we explained earlier, with a conspiracy charge, "the precise nature and extent of the conspiracy must be determined by reference to the agreement which embraces and defines its objects." Braverman, 317 U.S. at 53, 63 S.Ct. at 101. Through much use, we have found the Marable analysis particularly suited to defining the agreement. Since we are satisfied that changing the substantive test would not ease any of the complications with which we are concerned, we therefore have no reason not to apply Marable in this new context.

c. Sources of Proof

The question of what information a court may look to when its evaluation of the Marable factors follows a guilty plea is more difficult to resolve. To answer it requires some background information. In 1974, the United States Supreme Court decided Blackledge v. Perry, 417 U.S. 21, 94 S.Ct. 2098, 40 L.Ed.2d 628 (1974). The question which faced the Court in Blackledge was whether a petitioner's guilty plea to a charged felony waived his right to challenge the felony on due process grounds in a subsequent habeas proceeding. The Court determined that it did not, because the petitioner was, in that case, asserting his "right not to be haled into court at all upon the felony charge." Id. at 30, 94 S.Ct. at 2103. As the Court saw it, a petitioner does not waive his claim by pleading guilty when "[t]he very initiation of the proceedings against him ... operated to deny him due process of law." Id. at 30-31, 94 S.Ct. at 2103-04. One year later, the Supreme Court relied on its reasoning in Blackledge to permit a defendant to assert a double jeopardy claim on appeal despite the fact that he earlier pled guilty to the crime which he claimed placed him in double jeopardy. Menna v. New York, 423 U.S. 61, 62-63, 96 S.Ct. 241, 242, 46 L.Ed.2d 195 (1975) (per curiam). In a three page per curiam opinion, decided without oral argument, the Court held that, "[w]here the State is precluded by the United States Constitution from haling a defendant into court on a charge, federal law requires that a conviction on that charge be set aside even if the conviction was entered pursuant to a counseled plea of guilty." Id. at 62, 96 S.Ct. at 242. In a footnote, the Court explained its reasoning:

Id. at 62-63 n. 2, 96 S.Ct. at 242 n. 2. Instead of concluding on that note, however, the Supreme Court added a final, enigmatic comment:

Id. (emphasis added).

The question this final comment raises is whether by it, the Supreme Court meant also to decide its negative: that unless a double jeopardy claim is apparent when "judged on its face," the defendant does waive his claim by pleading guilty. If the Supreme Court meant to so hold, it would follow that a court entertaining a double jeopardy claim after a plea of guilty would necessarily be limited to the indictment or other charging instrument when reviewing the claim.8 We have not, however, so far examined the face of the indictments in a particular case to see if the defendant, after pleading guilty, retained the power to plead double jeopardy; instead, we have consistently read Menna as holding broadly that "the entry of a guilty plea does not waive a challenge based on a violation of the double jeopardy clause." United States v. Broussard, 645 F.2d 504, 505 (5th Cir. Unit A 1981); see also Long v. McCotter, 792 F.2d 1338, 1344 (5th Cir.1986) ("Menna ... holds that, when the constitutional claim asserted relates to the power of the court to try the defendant, it is not the type of claim that is waived."); Green v. Estelle, 601 F.2d 877, 877-78 (5th Cir.1979) ("[I]n Menna ... the Court held that a guilty plea to an offense does not bar a double jeopardy claim."); Green v. Estelle, 524 F.2d 1243, 1244 (5th Cir.1975) ("Menna ... holds that where a valid double jeopardy claim would preclude a constitutional prosecution, a plea of guilty to that charge does not extinguish the double jeopardy claim.").9 We do not believe that the Supreme Court would rely on a negative implication to explain that a defendant waives his constitutional right not to be placed in double jeopardy whenever he pleads guilty to an indictment which does not as a matter of law prove his claim. The clear and unyielding language the Court used in the body of the opinion — "federal law requires that a conviction ... be set aside even if the conviction was entered pursuant to a counseled plea of guilty" — supports our conclusion. Given the fact that Menna was decided per curiam and without oral argument, we think it more likely that the Supreme Court used the "judged on its face" language simply to describe how clear the double jeopardy issue before it was.10

This alternative explanation of Menna's troubling language appears more than reasonable given the nature of an indictment. The Supreme Court has repeatedly recognized that the protections which an indictment is intended to guarantee are reflected in the two criteria by which the Court measures an indictment's sufficiency: (1) whether the indictment "contains the elements of the offense intended to be charged, and sufficiently apprises the defendant of what he must be prepared to meet" and (2) "in case any other proceedings are taken against him for a similar offence, whether the record shows with accuracy to what extent he may plead a former acquittal or conviction." Russell v. United States, 369 U.S. 749, 764, 82 S.Ct. 1038, 1047, 8 L.Ed.2d 240 (1962) (quotations omitted) (citing cases). A "sufficient" indictment, therefore, will always provide a defendant with enough information to enable him to plead a double jeopardy claim. That does not mean, however, that an indictment is "insufficient" if it fails to present the information in enough detail to permit a defendant to prove, from the indictment, his double jeopardy claim. In fact, by explaining that a defendant can "rely upon other parts of the present record in the event that future proceedings should be taken against [him]," the Court in Russell acknowledged that an indictment's usefulness for proving a double jeopardy claim is limited. Russell, 369 U.S. at 764, 82 S.Ct. at 1047. Moreover, as long ago as 1913, the Supreme Court remarked that:

Bartell v. United States, 227 U.S. 427, 433, 33 S.Ct. 383, 385, 57 L.Ed. 583 (1913). Accord United States v. Lavergne, 805 F.2d 517, 521 (5th Cir.1986); United States v. Gordon, 780 F.2d 1165, 1172 (5th Cir.1986); United States v. Haas, 583 F.2d 216, 221 (5th Cir.1978), cert. denied, 440 U.S. 981, 99 S.Ct. 1788, 60 L.Ed.2d 240 (1979). When it formulated the standards by which indictments are to be measured, therefore, the Supreme Court assumed that a defendant, when necessary, could use information beyond that contained in the indictment to prove his double jeopardy claim. In other words, the Court recognized that indictments are often inadequate to prove double jeopardy. We reached this same conclusion in Marable when we explained that, "by the nature of the crime, the precise bounds of a single conspiracy seldom will be clear from the indictment." Marable, 578 F.2d at 153. However, if the "judged on its face" language of Menna is read to restrict a defendant to his indictment for proof of a double jeopardy claim once he has pled guilty to the crime he is challenging, we will necessarily be requiring an indictment to serve a purpose for which it is, admittedly, frequently inadequate. We might be less concerned about this result if it did not have constitutional, as well as practical, implications. As it is, however, we simply cannot, on the strength of this single, ambiguous phrase found in the footnote of a per curiam Supreme Court opinion, justify such a result. Consequently, we hold that Menna does not restrict a defendant's ability to raise a double jeopardy claim after a guilty plea to cases where the double jeopardy violation is clear from the face of the indictment. Similarly, therefore, we hold that Menna does not require us to look only to the indictment when reviewing a double jeopardy claim which follows a guilty plea.

By holding that Menna does not limit a court's examination to the indictment, we do not imply that it will always be necessary, or even appropriate, to look farther. As our cases demonstrate, our concern is caused by the freedom which the government has in drafting an indictment.11 See, e.g., Levy, 803 F.2d at 1395 (recognizing that the government "carve[d] out a portion of a larger conspiracy and allege[d] it as a separate conspiracy"); Dunn, 775 F.2d at 607 (expressing concern that the government could "separate a single, ongoing conspiracy into discrete conspiracies by simply dividing the time frame"); Winship (recognizing that "we cannot rest a conclusion of separate conspiracy personnel on the government's decision of who to charge in specific indictments"). Through the years, we have seen this freedom manifest itself in indictments which vaguely refer to agreements "beginning on or before" or "continuing from" a certain time, made in one or two named jurisdictions and "elsewhere," between the person indicted and "other persons to the grand jury known and unknown." Although we understand the importance, from both the government's and society's perspective, of permitting broad indictments, we also recognize the dangers of such indictments. As the Fourth Circuit eloquently explained:

Short v. United States, 91 F.2d 614, 624 (4th Cir.1937); accord Broce, 781 F.2d at 799, 804 (McKay, J., concurring) (Seymour, J., concurring in part and dissenting in part). When the government fills an indictment with vague allegations, therefore, our concern is at its greatest. However, if the government instead pleads the indictment with particularity, the need for our concern is less. And finally, when a detailed indictment is followed by a guilty plea, our concern can diminish to a point where we will not feel compelled to look beyond the indictment at all. The reason was given in Menna. As the Supreme Court explained, a guilty plea "validly removes the issue of factual guilt from the case." Menna, 423 U.S. at 62-63 n. 2, 96 S.Ct. at 242 n. 2. A petitioner who later contests his sentence on double jeopardy grounds after entering a guilty plea, therefore, has lost his right to challenge or dispute either the information in the indictment or any other facts he admitted when his guilty plea was taken. If the admitted facts, when viewed together, paint separate and distinct pictures which rule out the reasonable possibility of a double jeopardy problem, a court will be justified in relying solely on these facts. Resort to evidence outside the indictment, therefore, will be both unnecessary and inappropriate. See, e.g., United States v. Hall, 457 F.2d 1324, 1327-28 (5th Cir.1972).12 But the same cannot be true of indictments which depend on vague allegations to paint, with very broad strokes, only the barest suggestion of the crimes they charge. If the allegations in these vague indictments overlap in a way which suggests a double jeopardy problem, and evidence outside the record helps flesh-out the charges in the indictments, resort to that evidence is essential.

Turning specifically to the nature of the case before us, we find that a petitioner's decision to plead guilty requires us to modify our rule in Marable that, when drug conspiracy charges are involved, a court must look to more than the indictment when applying the Marable factors. As we explained above, once a defendant pleads guilty to a drug conspiracy charge, his ability to challenge the charge on double jeopardy grounds by reference to evidence outside the indictment will depend on the specificity with which the indictment was pled. If the faces of the indictments clearly prove a double jeopardy violation under the Marable five-factor test, of course, a court has no reason to resort to any other evidence. Similarly, if the indictments are pled so specifically that they rule out — on their faces — the reasonable possibility of a double jeopardy claim, a court should not examine evidence outside the indictment. However, in those cases which fall in between these two extremes, where vague allegations suggest the reasonable possibility of a double jeopardy violation but do not prove one, a court applying the Marable factors must consider evidence outside the indictments when it is offered.

3. Disposition of Atkins' Claim

Having set forth the standards we must apply, we are now ready to address the merits of Atkins' double jeopardy claim and the district court's judgment. We first note that both indictments under which Atkins pled guilty contain — when tested by the Marable factors — the sort of vague, overlapping allegations which require us to consider the additional evidence Atkins has offered. For example, the conspiracy charged in the Oklahoma indictment lasted from April, 1982 until April 13, 1983; the conspiracy charged in the Texas indictment began "on or before March 30, 1983" and ended April 12, 1983. Also, the Oklahoma indictment charged Atkins with conspiring with four individuals and "with diverse other persons whose names are to the Grand Jury unknown." The Texas indictment lists no co-conspirators at all; instead, it alluded only to "other persons to the grand jurors known and unknown." Both indictments charged Atkins with the same statutory offense — a 21 U.S.C. § 846 conspiracy to manufacture amphetamines in violation of 21 U.S.C. § 841(a)(1). The only difference between the two on this point is that the Oklahoma indictment also charged Atkins with conspiring to manufacture phenylacetone, the immediate precursor to amphetamine. Next, although the Oklahoma indictment lists four separate overt acts, the Texas indictment alleges none with which to compare it. Finally, the Oklahoma indictment places its conspiracy in the Western District of Oklahoma, the Northern District of Texas, and "elsewhere." The Texas indictment, although focusing primarily on the Western District of Texas, also suggests that its conspiracy extended to "divers other places to the grand jurors unknown." Neither indictment, therefore, expressly conflicts with the other on any of the points which the Marable test examines. Even on those points where they appear to differ, each indictment contains vague language which permits the inference that the additional people, places, or times from the other indictment could also be included. While the indictments do not disprove Atkins' claim, neither do they prove it. None of the facts on which the Marable factors focus matches exactly, although some come very close. Therefore, because the indictments affirmatively suggest the reasonable possibility that the same conspiracy was the focus of each, but fail to prove or disprove Atkins' double jeopardy claim, we must turn for help in evaluating Atkins' claim to the evidence he appended to his habeas petition.

When we look beyond the Texas and Oklahoma indictments, the distinctions which appeared on first application of the Marable factors begin to blur. The supporting evidence tends to infuse into the Texas indictment people and places with which the Oklahoma indictment is concerned. For example, the evidence suggests that Kenneth Gill's truck was found at Dansby's house in the Western District of Texas after the explosion — the site of the Texas indictment's conspiracy. Gill, of course, was charged as a co-conspirator in the Oklahoma indictment. Also, the Texas investigation appears to have only begun with the explosion at Dansby's house in the Western District of Texas on March 30, 1983 — the Texas indictment chooses to date the conspiracy from the explosion and conclude it two weeks later with Atkins' arrest in Oklahoma. However, all of the events which occurred after the explosion place Atkins in Dallas or Oklahoma — not the Western District of Texas. Finally, under the fourth Marable factor which requires us to look at "any other description of the offense charged which indicates the nature and scope of the activity which the government sought to punish in each case," the Texas and Oklahoma DEA investigations themselves are relevant. Although the investigations arguably appear to have been separate, they do not clearly appear to have been independent. The DEA reports can be read to suggest that the investigations shared key personnel and that the agents for the two investigations communicated with each other. Moreover, both Coonce — who is treated as an agent on the Texas investigation — and the Agent — who provided instrumental information to the Oklahoma investigation — seem to have been based in Dallas, Texas. Most importantly, the reports place both agents at the Metroplex Chemical Company in Dallas on the day Atkins bought chemicals and equipment, and DEA Special Agent Coonce is listed in the Oklahoma criminal complaint as a material witness to the Oklahoma investigation even though he was working on the Texas investigation.

As we explained initially, we must use this evidence to test the two implicit conclusions which the district court made in rejecting Atkins' claim. After doing so, we conclude first that the court's implicit finding that Atkins has not yet proved his double jeopardy claim is correct. While the evidence before us can be read to suggest an ongoing conspiracy which began in Texas and moved to Oklahoma after the explosion of the lab, it is not yet sufficient to prove the existence of only a single agreement. We believe that the details surrounding Atkins' manufacture of amphetamine in Texas, including who his co-conspirators were and how long he operated in Texas, need to be further explored before it can be said with certainty how many conspiracies Atkins engaged in. However, while the evidence Atkins presented is not enough to prove his claim, it is certainly sufficient under the standards set forth in section 2255 to entitle Atkins to an evidentiary hearing on his claim. We find, therefore, that the district court erred by not convening a hearing and, therefore, by implicitly concluding that "the motion and the files and records of the case conclusively show[ed] that the prisoner is entitled to no relief." See 28 U.S.C. § 2255. We, therefore, must remand Atkins' claim to the district court.

We make one final observation for the benefit of the district court on remand. Throughout this opinion, we have been guided by the accepted principle that a petitioner who raises a double jeopardy claim in the manner Atkins' claim has been raised must assume responsibility, within permissible boundaries, for the problems he causes by raising the claim first in a habeas proceeding. In accordance with that principle, we have recognized that the petitioner bears the burden of proving double jeopardy. We have also recognized that because he admits particular facts by pleading guilty, a petitioner cannot prove his claim by relying on evidence which contradicts those facts. Finally, and related to the last point, we have recognized that if the indictments charging the petitioner are sufficiently detailed and specific, and rule out the reasonable possibility of a double jeopardy violation, the petitioner will not be allowed to present any evidence outside of that contained in the indictments.

Atkins did not just plead guilty to the Texas indictment, however; his guilty plea was the result of a plea bargain pursuant to which the government dismissed count two of the Texas indictment. Count two charged Atkins with illegally manufacturing twelve ounces of amphetamine — a crime separate and distinct from either of the conspiracy charges we have examined today. See Levy, 803 F.2d at 1397 (the double jeopardy clause does not preclude the government from prosecuting a defendant for both the conspiracy to commit a crime and the crime itself). If, on remand, Atkins is successful in proving his double jeopardy claim and his sentence is vacated, by his actions he will have tacitly repudiated the plea bargain. See Moore v. Foti, 546 F.2d 67 (5th Cir.1977) (a defendant's successful challenge to his plea-bargained sentence is a tacit repudiation of the bargain). Recently, in Fransaw v. Lynaugh, 810 F.2d 518 (5th Cir.), cert. denied, ___ U.S. ___, 107 S.Ct. 3237, 97 L.Ed.2d 742 (1987), we had occasion to examine in detail the effect of a repudiated plea bargain on the charges dismissed as part of that bargain. We recognized that "[t]he cases hold with apparent unanimity that when defendant repudiates the plea bargain ... there is no double jeopardy (or other) obstacle to restoring the relationship between defendant and state as it existed prior to the defunct bargain." Id. at 524-25. We instruct the district court, therefore, that should Atkins' claim be successful on remand, it would be appropriate for it to reinstate count two of the Texas indictment which the government dismissed as part of its plea bargain with Atkins.

III.

For the above described reasons, the judgment of the district court is REVERSED and Atkins' claim is REMANDED to that court for an evidentiary hearing.